[double cursor]
prlw1@newn.cam.ac.uk said:
> Just to make it clear - this is just with pcdisplay, not with vga 

Yes - and Andy has identified the bug and will fix it.

[your patch]
That's exactly what I wanted to tell.
I forgot one more before the "dispoffset" however: can you comment
out everything between the
	 * check if this is really a VGA
and - not including -
	rv = 1;
in vga_common_probe()? Perhaps this does some harm to the card.

The "dispoffset": You could replace the
		scr->pcs.dispoffset = vga_6845_read(&vc->hdl, startadrh) << 9;
		scr->pcs.dispoffset |= vga_6845_read(&vc->hdl, startadrl) << 1;
by a mere
		scr->pcs.dispoffset = 0;
as a start.
As long as vga_copyrows() isn't used, nothing should touch the display offset
register anymore.

Now I don't see anything which "vga" does more than "pcdisplay"
anymore. (As long as no font loads or screen switches are done.)

best regards
Matthias